In August 2021, Intel first proposed a plan to build a "super chip factory" in the United States. It plans to invest US$100 billion and employ tens of thousands of people to manufacture chips with billions of transistor integration scales. chip. In the same month, 22-year-old student Sam Zeloof also announced the results of his own homemade chip in the garage of his home in New Jersey, USA.
According to Wire magazine, Zeloof used collected second-hand equipment and homemade equipment to create an integrated circuit with 1,200 transistors, code-named Z2. He cut the silicon wafers himself, etched the circuits with ultraviolet light, and then soaked them in sulfuric acid, a conventional but "primitive" way to create the chips.
He said, "Maybe I am overconfident, but I believe that if others have found a way, then I can do it, even if it takes longer,"
The Z2 is the second chip produced by Zeloof. The first chip, the Z1, was smaller and came out in 2018. In 2017, he began manufacturing individual transistors.
Although these products were played by Intel 50 years ago, Zeloof joked that his process progress is faster than that of the semiconductor industry in its infancy. The number of transistors in Z2 is 200 times that of Z1, far exceeding Moore's Law (the industry law that the number of transistors per chip doubles every 18 months)
Zeloof hopes to achieve the same code name as Intel's 1971 launch Same integration scale as 4004 processor. The 4004 was Intel's first commercial microprocessor launched in 1971, integrating 2,300 transistors. The 22-year-old student began working on a circuit design that could perform simple addition operations a month ago.
He said that chips made in garages are unlikely to appear in high-end electronic products, but he believes that the results achieved by his somewhat alternative interests can also show that inventors do not have a million-dollar budget. Chip manufacturing technology will be available, and society will benefit from it.
Not only laymen but also industry insiders believe that making chips is the most difficult and expensive project in the world. When Zeloof initially told several experts in the semiconductor industry about his project, they emailed him back that it was impossible.
He later received support from his family, but their attitude was also more cautious. His father once asked a semiconductor engineer named Mark Rothman, who has 40 years of experience in semiconductor engineering, for advice. "My first reaction was that he couldn't do it. It was just a garage. But when I noticed that the young man As the work progressed, my mind slowly changed. He did things I never thought I could do personally." Zeloof had to read a lot of historical and technical material for his project. Today's semiconductor production lines are in giant modern factories. These facilities are also equipped with air conditioning and dust-free ventilation systems that can work continuously for a long time to prevent the silicon substrates from being contaminated with dust. Zeloof couldn't do that, so he had to read patent specifications and textbooks from the 1960s and 1970s, when engineers were still making chips at regular desks.
He also purchased some equipment as well as obsolete equipment. His most expensive piece of equipment was a broken electron microscope that he bought for $1,000 and repaired, using it to find defects in chips. You know, in the early 1990s, it cost $250,000.
Zeloof had to etch his designed circuits onto a silicon substrate, but he couldn't afford a $150 million photolithography machine. He eventually found his own solution, which involved connecting a conference projector to an electron microscope, which would project the circuit diagram onto a silicon substrate coated with a photosensitive material.
His first-generation product, the Z1, used a 175-micron process and integrated six transistors.
The Z2 chip integrates 1,200 transistors, the computing speed is increased by 10 times, and the process is upgraded to 10 microns.
Zeloof recently upgraded his homemade photolithography machine, which is capable of etching circuits of 0.3 microns, or 300 nanometers, which is equivalent to the level of semiconductor technology in the mid-1990s. He is currently considering implementing a computing chip with the same integration scale and performance as Intel's 4004 series.
He said, "I want to upgrade the homemade chip technology and let everyone see what we can do at home."